Why Lidar
From Points to Products Business Benefits from Lidar using ArcGIS 10.1 Functionality Mark Norris-Rogers, Mondi Ltd. Ron Behrendt, Behron LLC
Mondi at a Glance -
International vertically integrated
forestry, pulp and paper company -
Mondi South Africa: o 306 000 ha: o 4 regions in KwaZuluNatal & o Mpumalanga o 202
000 ha: Planted o Eucalyptus: 145 000 ha o Pine: 41 000ha o Wattle: 16 000 ha
o Ownership o Owned: 227 000 ha o Leased: 78 500 ha o Managed: 500 ha
Pulpwood Species
4 year clonal stand of Pinus elliottii x P. caribaea
Wattle
Pine 26 March 2013 Pine/CTE research nursery PAGE 9
Eucalyptus
Mondi’s GIS History o 1994-2001: Arc/Info (Unix); ArcSTORM; o Distributed databases – little integration o One question = Different Answers.
ArcView
o 2001-2011: Customised MapObjects App o Centralised SDE Database o Edits on thick client via Citrix – Very slow o Line drops = data corruption
o 2011: ArcGIS 10.0 Replicated GDB o 2-Way replication – Regions edit ; replicate up; synch PMi o Overcome network constraints o Maintain advantage of centralised DB; work locally
Why Lidar ? o Forestry
fundamentally spatial in nature
o
Areas (ha/ac); Distance (km/mi); Elevation/Height (m/ft); Yield (m3/ha; board feet/ac)
o
Not only horizontal x/y axes but vertical z axis as well
o
Move from 2D to 3D
o Terrain o
plays a critical role
Need to derive accurate Digital Elevation Models (DEMs) that accurately describe terrain: o Elevation; Slope; Aspect; Ground Roughness, etc. o Terrain impacts site classification; sustainability; forest operations; machine access; harvest planning o Difficult to derive accurate terrain data when covered by forest – usually derive a surface model (i.e. top of canopy)
Harvesters on Flat to Steep Terrain
Why Lidar ? o Harvest
Planning
o
Accessibility of mechanized equipment into forest stands
o
High road banks or berms can hinder access
o
Identification of sites where timber extracted from a forest stand can be stacked
o Forest
Stand Structure
o
Tree heights; Basal Area; Stocking (Stems per hectare/acre); Canopy Diameter; Tree/Stand Volume estimates
o
Biomass estimation
o Lidar
3D
provides appropriate data to derive this information – inherently
Mechanized Harvesting Equipment
Lidar Acquisition – Step 1
Questions to answer: o What
should we order?
o Who
should we purchase it from?
o How
much will it cost?
Lidar Acquisition – Critical Parameters oUnderstand what minimum criteria are required to achieve objectives: o Geographic extent of area to be surveyed (supply shapefile) o Minimum point density – forestry ~ 6 points/m2 ; unless only require Bare Earth surface ~2-4 points/m2 o Field of View – should not exceed 15º off nadir (i.e. max 30º). Risk of poor data if wider. o Overlapping strips – 30-50% dependant on sensor/platform. o Point Classification – Ground/Non-ground as a minimum o Classify overlap points according to return, not as overlap.
Lidar Acquisition – Critical Parameters o Understand
what minimum criteria are required to achieve
objectives: o
Ground Control Points – sufficient to guarantee required spatial accuracy
o
Data processing – according to industry standards
o
Lidar delivered as tiled .LAS format point cloud
o
Orthophoto imagery – Should be acquired simultaneously, with a 15cm or better ground resolution
o
Specify if require Natural Color imagery or (preferably) Near Infrared band (False Color IR Imagery)
o
Cloud/Smoke free imagery
o
Specify Projection Parameters for delivered data
Lidar Acquisition – Vendor Selection •
Lessons Learned:
o
Make certain vendors understand what YOU require
o
Provide as comprehensive a specification as possible, but listen to their suggestions
o
Meet with preferred Vendor/s prior to signing any contract – Discuss specs and options. Can save costs o
Various approaches to mobile GPS corrections (i.e. Precision Point Positioning vs GPS base stations)
o
Required accuracy: be realistic sub centimeter vs. 1 meter, allowed a reduction in the number of control points
o
USD $5,000 saving
o
Vendors tend to have standard response document for Tenders/RFPs – just adjust the costs
o
Order the products you require and not what the vendor thinks you want!
Lidar Acquisition – On Receiving Data o Ingesting o
Lidar data into ArcGIS 10.1 LAS Datasets (.LASD)
ArcGIS 10.1 provides a new data format specifically for Lidar Data –
LAS Dataset (.lasd file extension) o
Enables ArcGIS 10.1 to work with Lidar files in their native .las format
o
Depending on data volume, might need to work with several LAS Datasets to cover full Lidar area
Lidar Acquisition – On Receiving Data o
Quality Control of Lidar Data o
Necessary to review quality of data received – Are specs met?
o
minimum points/m2?;
o
outliers? (points below ground level or far above maximum heights – bird strikes etc);
o
Gaps in point coverage, especially ground returns;
o
Can use LAS Point Statistics as Raster tool in ArcGIS 10.1 to create QC layers
o
Missing Data
Outliers
Creating Base Products o Next
Step – Create Derivative (Raster) Products
o
Digital Elevation Models (DEMs) - Bare earth surfaces; Contours; Hillshades
o
Digital Surface Models (DSMs) – Top-of-Canopy/Buildings surfaces; Hillshades
o
Slope and Slope Class Surfaces
o
Aspect Surfaces
o
Canopy Height Models (CHMs) – Tree height surfaces
o
Canopy Density Models (CDMs) – Stocking/survival assessment surfaces
o Notes:
Vendors can usually supply all/some of these products, but preferable to derive one’s own products
o
o
Can better understand the data and extract more information out of it - vendors don't know your environment
o
All products were created using out-of-the box ArcGIS 10.1 functionality
Lidar unaffected by Shadow
Orthophoto DEM DSM
CHM
Value Add Product Applications Terrain Visualization; Ground Roughness False Colour Infrared Orthophoto
Lidar Derived Surface Model - 1m Grid Note Ground Roughness
Value Add Product Applications Terrain Visualization; Slope Class Data Current Slope Class Data - 20m Grid derived from Dot-Roll Contour Method off Stereoplotter
Lidar Derived Slope Class Data - 1m Grid Note increase in 45-60; >60% slopes
• Slopes from 45-60% require self-leveling harvesters (higher cost than normal harvesters). • Calculate accurate areas of these slopes, where previously only estimated areas were available • Minimizes accident risk due to incorrect machine access.
Value Add Product Applications Machine Access and Extraction Route Planning Identification of Road Banks (red strips parallel to roads) – Stand Access Limitations
Slope (Percent)
Improved Contour; Road Alignment Data Current 10m Contours (in blue) vs. Lidar derived 2m Contours (in red)
Road Alignment: GIS Data vs. Lidar Data
Road Alignment Data – Why? Mondi Logging Truck
Canopy Height Model–Tree Height Metrics False Colour Infrared Orthophoto
Lidar Canopy Height Model - 1m Grid Height in Metres
Improved Stand Delineation - CHM Improved Stand Boundary Delineation using the Canopy Height Model data
Identification of Unfelled Patches using the Canopy Height Model data
Improved Stand Delineation - DSM Improved Stand Boundary Delineation using the Canopy Height Model data
Improved Stand Boundary Delineation using the DSM Hillshade Model data
More Detailed Aspect Data Aspect Data derived from 20m DEM derived from Dot-Roll Contour Method off Stereoplotter
Aspect Data derived from 1m Lidar DEM
Digital Surface Model – Stand Uniformity False Colour Infrared Orthophoto
Lidar Canopy Height Model - 1m Grid Hillshade- Indication of Canopy Uniformity
Value Add Product Applications o Environmental Applications
o
Erosion ditch identification
o
Landscape-level land use planning
o
High Conservation Value Forest identification
o
Invasive Alien Plant mapping
Environmental Applications Erosion Channels highlighted in DEM Hillshade
High Conservation Value Riverine Forest Patch
Conclusions o Concerns
o
o
o
with Lidar – Cost; Data Volume
o
Value of Lidar is in the level of detail gained, and
o
multiple products that currently require multiple systems to obtain.
o
Some of the products can only be obtained by Lidar
Impact of Slope Class Definition – o
Ability to calculate exact areas of specific slope classes:
o
Improved Harvest Planning, especially for costly harvesters
o
Reduced accident risk – Accurate machine/terrain matching
Identification of Access/Extraction problems o
e.g. due to road banks – hinder access; log storage/loading decks
o
Improved planning for stand access points; landing zones;
Improved Data Maintenance o
Improved road delineation (more accurate road network; distance, road maintenance calculations)
o
Improved stand delineation (improved area; yield calculations)
Future Plans o Phase
2 Goals (2013)
o
Quantitative Forest Stand metrics – stocking(stems per hectare); basal area; DBH classes;
o
Timber Volume Estimates – Standing volumes; Average tree size
o
Automated Road Extraction
o Longer o
Term (3-5 years) – Integrated Forest Monitoring Program
Current Forest Monitoring Program:
Annual multispectral image acquisition program (~120 000 ha/yr) Ground-based plot sampling enumeration program – forest inventory measurements Various auditing/ad-hoc checking programs to check planting survival; database accuracies etc. o
Plan to replace all these programs with a Lidar-based monitoring program
Entire forest base will be surveyed on a two-year cycle. Estimated saving of 20-25% of current monitoring program costs
Acknowledgments: •
Cody Benkelman, Imagery Product Manager, Esri Inc. for his technical guidance and input;
•
Peter Eredics, Forestry Manager, Esri Inc. for his approval of and support for this project;
•
•
Mondi Ltd, for making the Lidar data available. You - for listening!
Thank you
Mondi at a Glance -
International vertically integrated
forestry, pulp and paper company -
Mondi South Africa: o 306 000 ha: o 4 regions in KwaZuluNatal & o Mpumalanga o 202
000 ha: Planted o Eucalyptus: 145 000 ha o Pine: 41 000ha o Wattle: 16 000 ha
o Ownership o Owned: 227 000 ha o Leased: 78 500 ha o Managed: 500 ha
Pulpwood Species
4 year clonal stand of Pinus elliottii x P. caribaea
Wattle
Pine 26 March 2013 Pine/CTE research nursery PAGE 9
Eucalyptus
Mondi’s GIS History o 1994-2001: Arc/Info (Unix); ArcSTORM; o Distributed databases – little integration o One question = Different Answers.
ArcView
o 2001-2011: Customised MapObjects App o Centralised SDE Database o Edits on thick client via Citrix – Very slow o Line drops = data corruption
o 2011: ArcGIS 10.0 Replicated GDB o 2-Way replication – Regions edit ; replicate up; synch PMi o Overcome network constraints o Maintain advantage of centralised DB; work locally
Why Lidar ? o Forestry
fundamentally spatial in nature
o
Areas (ha/ac); Distance (km/mi); Elevation/Height (m/ft); Yield (m3/ha; board feet/ac)
o
Not only horizontal x/y axes but vertical z axis as well
o
Move from 2D to 3D
o Terrain o
plays a critical role
Need to derive accurate Digital Elevation Models (DEMs) that accurately describe terrain: o Elevation; Slope; Aspect; Ground Roughness, etc. o Terrain impacts site classification; sustainability; forest operations; machine access; harvest planning o Difficult to derive accurate terrain data when covered by forest – usually derive a surface model (i.e. top of canopy)
Harvesters on Flat to Steep Terrain
Why Lidar ? o Harvest
Planning
o
Accessibility of mechanized equipment into forest stands
o
High road banks or berms can hinder access
o
Identification of sites where timber extracted from a forest stand can be stacked
o Forest
Stand Structure
o
Tree heights; Basal Area; Stocking (Stems per hectare/acre); Canopy Diameter; Tree/Stand Volume estimates
o
Biomass estimation
o Lidar
3D
provides appropriate data to derive this information – inherently
Mechanized Harvesting Equipment
Lidar Acquisition – Step 1
Questions to answer: o What
should we order?
o Who
should we purchase it from?
o How
much will it cost?
Lidar Acquisition – Critical Parameters oUnderstand what minimum criteria are required to achieve objectives: o Geographic extent of area to be surveyed (supply shapefile) o Minimum point density – forestry ~ 6 points/m2 ; unless only require Bare Earth surface ~2-4 points/m2 o Field of View – should not exceed 15º off nadir (i.e. max 30º). Risk of poor data if wider. o Overlapping strips – 30-50% dependant on sensor/platform. o Point Classification – Ground/Non-ground as a minimum o Classify overlap points according to return, not as overlap.
Lidar Acquisition – Critical Parameters o Understand
what minimum criteria are required to achieve
objectives: o
Ground Control Points – sufficient to guarantee required spatial accuracy
o
Data processing – according to industry standards
o
Lidar delivered as tiled .LAS format point cloud
o
Orthophoto imagery – Should be acquired simultaneously, with a 15cm or better ground resolution
o
Specify if require Natural Color imagery or (preferably) Near Infrared band (False Color IR Imagery)
o
Cloud/Smoke free imagery
o
Specify Projection Parameters for delivered data
Lidar Acquisition – Vendor Selection •
Lessons Learned:
o
Make certain vendors understand what YOU require
o
Provide as comprehensive a specification as possible, but listen to their suggestions
o
Meet with preferred Vendor/s prior to signing any contract – Discuss specs and options. Can save costs o
Various approaches to mobile GPS corrections (i.e. Precision Point Positioning vs GPS base stations)
o
Required accuracy: be realistic sub centimeter vs. 1 meter, allowed a reduction in the number of control points
o
USD $5,000 saving
o
Vendors tend to have standard response document for Tenders/RFPs – just adjust the costs
o
Order the products you require and not what the vendor thinks you want!
Lidar Acquisition – On Receiving Data o Ingesting o
Lidar data into ArcGIS 10.1 LAS Datasets (.LASD)
ArcGIS 10.1 provides a new data format specifically for Lidar Data –
LAS Dataset (.lasd file extension) o
Enables ArcGIS 10.1 to work with Lidar files in their native .las format
o
Depending on data volume, might need to work with several LAS Datasets to cover full Lidar area
Lidar Acquisition – On Receiving Data o
Quality Control of Lidar Data o
Necessary to review quality of data received – Are specs met?
o
minimum points/m2?;
o
outliers? (points below ground level or far above maximum heights – bird strikes etc);
o
Gaps in point coverage, especially ground returns;
o
Can use LAS Point Statistics as Raster tool in ArcGIS 10.1 to create QC layers
o
Missing Data
Outliers
Creating Base Products o Next
Step – Create Derivative (Raster) Products
o
Digital Elevation Models (DEMs) - Bare earth surfaces; Contours; Hillshades
o
Digital Surface Models (DSMs) – Top-of-Canopy/Buildings surfaces; Hillshades
o
Slope and Slope Class Surfaces
o
Aspect Surfaces
o
Canopy Height Models (CHMs) – Tree height surfaces
o
Canopy Density Models (CDMs) – Stocking/survival assessment surfaces
o Notes:
Vendors can usually supply all/some of these products, but preferable to derive one’s own products
o
o
Can better understand the data and extract more information out of it - vendors don't know your environment
o
All products were created using out-of-the box ArcGIS 10.1 functionality
Lidar unaffected by Shadow
Orthophoto DEM DSM
CHM
Value Add Product Applications Terrain Visualization; Ground Roughness False Colour Infrared Orthophoto
Lidar Derived Surface Model - 1m Grid Note Ground Roughness
Value Add Product Applications Terrain Visualization; Slope Class Data Current Slope Class Data - 20m Grid derived from Dot-Roll Contour Method off Stereoplotter
Lidar Derived Slope Class Data - 1m Grid Note increase in 45-60; >60% slopes
• Slopes from 45-60% require self-leveling harvesters (higher cost than normal harvesters). • Calculate accurate areas of these slopes, where previously only estimated areas were available • Minimizes accident risk due to incorrect machine access.
Value Add Product Applications Machine Access and Extraction Route Planning Identification of Road Banks (red strips parallel to roads) – Stand Access Limitations
Slope (Percent)
Improved Contour; Road Alignment Data Current 10m Contours (in blue) vs. Lidar derived 2m Contours (in red)
Road Alignment: GIS Data vs. Lidar Data
Road Alignment Data – Why? Mondi Logging Truck
Canopy Height Model–Tree Height Metrics False Colour Infrared Orthophoto
Lidar Canopy Height Model - 1m Grid Height in Metres
Improved Stand Delineation - CHM Improved Stand Boundary Delineation using the Canopy Height Model data
Identification of Unfelled Patches using the Canopy Height Model data
Improved Stand Delineation - DSM Improved Stand Boundary Delineation using the Canopy Height Model data
Improved Stand Boundary Delineation using the DSM Hillshade Model data
More Detailed Aspect Data Aspect Data derived from 20m DEM derived from Dot-Roll Contour Method off Stereoplotter
Aspect Data derived from 1m Lidar DEM
Digital Surface Model – Stand Uniformity False Colour Infrared Orthophoto
Lidar Canopy Height Model - 1m Grid Hillshade- Indication of Canopy Uniformity
Value Add Product Applications o Environmental Applications
o
Erosion ditch identification
o
Landscape-level land use planning
o
High Conservation Value Forest identification
o
Invasive Alien Plant mapping
Environmental Applications Erosion Channels highlighted in DEM Hillshade
High Conservation Value Riverine Forest Patch
Conclusions o Concerns
o
o
o
with Lidar – Cost; Data Volume
o
Value of Lidar is in the level of detail gained, and
o
multiple products that currently require multiple systems to obtain.
o
Some of the products can only be obtained by Lidar
Impact of Slope Class Definition – o
Ability to calculate exact areas of specific slope classes:
o
Improved Harvest Planning, especially for costly harvesters
o
Reduced accident risk – Accurate machine/terrain matching
Identification of Access/Extraction problems o
e.g. due to road banks – hinder access; log storage/loading decks
o
Improved planning for stand access points; landing zones;
Improved Data Maintenance o
Improved road delineation (more accurate road network; distance, road maintenance calculations)
o
Improved stand delineation (improved area; yield calculations)
Future Plans o Phase
2 Goals (2013)
o
Quantitative Forest Stand metrics – stocking(stems per hectare); basal area; DBH classes;
o
Timber Volume Estimates – Standing volumes; Average tree size
o
Automated Road Extraction
o Longer o
Term (3-5 years) – Integrated Forest Monitoring Program
Current Forest Monitoring Program:
Annual multispectral image acquisition program (~120 000 ha/yr) Ground-based plot sampling enumeration program – forest inventory measurements Various auditing/ad-hoc checking programs to check planting survival; database accuracies etc. o
Plan to replace all these programs with a Lidar-based monitoring program
Entire forest base will be surveyed on a two-year cycle. Estimated saving of 20-25% of current monitoring program costs
Acknowledgments: •
Cody Benkelman, Imagery Product Manager, Esri Inc. for his technical guidance and input;
•
Peter Eredics, Forestry Manager, Esri Inc. for his approval of and support for this project;
•
•
Mondi Ltd, for making the Lidar data available. You - for listening!
Thank you
